K-3D is still quite early in development, however, there are some points that should be made about it’s status. The initial work up to 0.4 (which is now the stable branch) was geared towards providing a powerful, stable core from which to work. This aim was achieved incredibly well, and while the UI was to some degree neglected during this phase, it is still quite usable.
Current development efforts on a separate branch are to put together a completely new user interface that has been defined and designed by users for users. Work moves ahead at quite a pace on this, again due to the solid foundations laid during the initial phases.
Bottom line, while functionally K-3D cannot compare at the moment with Blender, it is (I am prepared to bet) going to take shape very fast, and in ways that Blender cannot easily achieve. If anyone here is programming oriented, I’d recommend looking at the source code to K-3D, it is a completely different situation to looking at Blender’s source, which is quite scary, even to an experienced coder. K-3D has a bright future, simply due to the fact that it has been engineered well from day one.
On the subject of awsome [sic] features that K-3D has, well, someone has already mentioned the macro recording feature, used to produce the tutorials, which is amazing in it’s own right. Everything is animatable, by the nature of the underlying system. System wide, multi-level undo. Seamless and powerful integration with Renderman renderers, and Yafray for those that need it.